Kubernetes (K8s) provides the foundation for integrating distributed edge-cloud resources. However, existing frameworks struggle to address the challenges of cross-cluster coordination and dynamic resource changes, limiting throughput. We propose KMPS, a deep reinforcement learning-based scheduling framework to enhance long-term throughput. KMPS integrates a multi-agent proximal policy optimization algorithm for autonomous edge access point scheduling, combined with gRPC cross-cluster scheduling and invalid target filtering; utilizes graph neural networks to embed system state information, decomposing high-dimensional service orchestration actions through multiple separate policy networks; and constructs a three-time-scale coordination mechanism (0.25s, 2s, 25s) to coordinate scheduling and orchestration, with K8s compatibility. Experiments on real workloads verify that KMPS operates stably under dynamic loads, sudden emergency tasks, and multi-cluster scenarios. Compared to baselines, the proposed framework achieves an over 5.3% increase in long-term throughput and a 60% reduction in cross-cluster scheduling latency.

Neural Architecture Search (NAS) requires global topological exploration and is hence time consuming. To address this challenge, we propose the comprehensive-forecast multiobjective genetic programming for NAS, or CFMOGP-NAS for short. By integrating the strengths of various regression models and synthesizing the forecast of multiple candidates, the accuracy and robustness of architecture predictions are enhanced. The resultant algorithm this way incorporates a strategy of a mixture of complete and partial training, which balances cost and accuracy of evaluation. To also balance the population diversity, we develop a regularized tournament scheme for genetic programming. Experimental studies show that CFMOGP-NAS achieves a 50% reduction in search time without sacrificing accuracy, and verify that it substantially improves efficacy and efficiency compared with the state-of-the-art NAS methods.

While there exist photovoltaic (PV) forecasting methods, persistent challenges remain, such as dynamically changing working conditions, nonlinearities, limited interpretability, and dependency on quality datasets. To address these issues, this article proposes a multiscale-dynamic and neural-tuned physics-formulated PV forecasting method, and thus develops a physics-embedded automated modeling (PEAM) framework for engineering systems. Different from a physics-informed neural network (PINN), PEAM explicitly integrates first-principle derived physical equations directly into the model for interpretability, with neural networks exploiting their nonlinear fitting and adaptation capabilities. Here, Bayesian optimization is utilized systematically to address hyperparameter sensitivity and significance analysis, thus automatically identifying critical environmental impact factors, focusing on the most influential input variables, and providing effacacy for data usage. Comprehensive experimental validation conducted on three different prediction steps, ten testing datasets, and one extreme weather test set from two real-world solar stations shows that PEAM consistently outperforms existing methods such as PINN. Overall, the PEAM framework enhances the predictive accuracy, interpretability, and robustness for practical deployment of PV forecasting systems.

Designing complex computer-aided design (CAD) models is often time-consuming due to challenges such as computational inefficiency and the difficulty of generating precise models. We propose a novel language-guided framework for industrial design automation to address these issues, integrating large language models (LLMs) with computer-automated design (CAutoD).Through this framework, CAD models are automatically generated from parameters and appearance descriptions, supporting the automation of design tasks during the detailed CAD design phase. Our approach introduces three key innovations: (1) a semi-automated data annotation pipeline that leverages LLMs and vision-language large models (VLLMs) to generate high-quality parameters and appearance descriptions; (2) a Transformer-based CAD generator (TCADGen) that predicts modeling sequences via dual-channel feature aggregation; (3) an enhanced CAD modeling generation model, called CADLLM, that is designed to refine the generated sequences by incorporating the confidence scores from TCADGen. Experimental results demonstrate that the proposed approach outperforms traditional methods in both accuracy and efficiency, providing a powerful tool for automating industrial workflows and generating complex CAD models from textual prompts.The code is available at https://jianxliao.github.io/cadllm-page/

Automated design of analog integrated circuits (ICs) involves balancing multiple objectives under process, voltage, and temperature (PVT) variations. An excess of constraints can ensnare algorithms in local optima, while the variations elevate the costs of simulation. To address this challenge, we propose a two-search mode multi-task evolutionary framework to balance objective optimization and constraint satisfaction under variations. Specifically, considering the inherent relationships between objective optimizations and constraint violations, our method adaptively switches between unconstrained surrogate-assisted and constrained simulation-driven search modes. Furthermore, our framework treats PVT variations as a multi-task challenge, facilitating inter-corner knowledge transfer via multi-task evolution, substantially lowering simulation costs. Our framework has been evaluated using two different sensing elements and an amplifier within a 22 nm process. Based on Monte-Carlo simulations, compared to multi-task reinforcement learning, this method attains a 60% to 80% reduction in the relative inaccuracy of sensing elements and accomplishes a 60% decrease in total runtime.

In the post-Moore era, chips integrate multiple technology node chiplets, necessitating repeated implementations of the same circuit topology across nodes, highlighting the need for technology-independent circuit representation. We use a four-parameter vector---gm, ft, VDS, and ΔVGS-to represent the behavior of each transistor, called the transistor behavioral vector (TBV). The TBVs are vertically concatenated to form the transistor behavioral circuit representation (TBCR) matrix, which precisely reflects the circuit's performance and provides a technology-independent representation. Furthermore, we propose a transistor behavioral model (TBM) to convert the TBV into the corresponding sizing. Finally, we propose a method to transfer analog circuits between different technology nodes using TBM (TNT), translating the modifications in the process parameters into the corresponding adjustments in ΔVGS. The experimental results show that for a single transistor, the mapping accuracy from TBV to simulation result was reached 99%. Multiple amplifiers were transferred from 180nm to 22nm technology, compared to the conventional transfer method based on gm/id, our transfer method based on TBCR achieved a success rate of up to 5× higher, along with additional performance improvements from the scaling down.

Analog IC design is mainly manual and implemented at the device level. A major reason is circuit behavior-extraction. Unlike its digital counterpart, analog IC design is strongly coupled with technology nodes and is difficult to represent by an abstract behavioral model. The lack of accurate and efficient analog modeling has become a bottleneck in analog design automation. This paper proposes a behavior-centric optimization framework for analog circuits that represents circuit behavior using transistor electrical properties instead of sizes, improving model generalization and reducing optimization complexity. To characterize the process, we propose a method for mapping transistor electrical properties to sizes. Moreover, we developed a radial basis functions-based Kolmogorov-Arnold network (RBF-KAN) to accurately approximate circuit nonlinear behavior with limited simulations. Compared to blackbox modeling, our approach enables constructing surrogate models via KAN under a set specification with just a few hundred simulations. Experiments on the testing suite showed our framework achieved a $1.76 \times$ to $2.64 \times$ improvement in large signal figure of merit (FOM) and $1.73 \times$ to $2.48 \times$ in small signal FOM over state-of-the-art methods, while also enabling $3.5 \times$ to $6.2 \times$ acceleration in design porting.

Clustered federated learning (FL), which groups clients with similar data distributions for collaborative training, represents a pivotal technique within federated learning for effectively addressing the challenges posed by non-IID data on clients. Existing clustered FL algorithms typically endeavor to learn distribution similarities of clients iteratively or indirectly through representations like gradients and loss. This necessitates resource-intensive pre-training or multiple iterations to attain stable clusters, thereby incurring additional communication cost and computational overhead. To address the above issues, we propose lightweight Clustered Federated Learning via Feature Extraction (FECFL). FECFL adopts a simple but effective client representation, i.e., the features extracted from the clients’ data using identically initialized models without any pre-training, to perform efficient one-shot clustering. Moreover, client data distribution is often dynamic in practice. To tackle distribution shift, we embed a distribution monitoring mechanism in FECFL, enabling adaptive re-grouping for new distributions. Finally, we demonstrate the benefits of FECFL over the baselines by conducting experiments on various datasets and distributions.

Retrieval-Augmented Generation (RAG) provides large language models (LLMs) a means of retrieving relevant external knowledge, but its document parsing leads to increased latency and energy consumption. To address this issue, we propose a dynamically reconfigurable Neural Processing Unit (NPU) that accelerates both RAG document parsing and inference. By leveraging compute-in-memory fusion, dynamic convolution, and multi-level parallelism, our approach reduces memory transfer overhead and optimizes hardware resource allocation. Experimental results show that our design achieves a 1.8x speedup in document parsing and a 2.83x improvement in energy efficiency. Additionally, it achieves an 11.71% reduction in inference time and a 35.59% boost in energy efficiency over traditional CPU/GPU methods, offering a scalable solution for large-scale RAG tasks.

At present, accurate detection of small objects in an aerial imagery remains a challenge in remote sensing due to limited pixel resolution, background clutter, and scale variations. To address these issues for high-precision detection in a complex remote sensing scene, we propose a novel detection framework based on RepViT Dynamic Fusion and YOLOv11, termed RDF-YOLO. The RDF-YOLO brings in two core innovations:(1) a Dynamic Scale RepViT module that integrates lightweight Transformer operations into the backbone to enhance global context modeling and semantic discrimination under noisy conditions, and (2) a dynamic fusion module that incorporates spatially aware dilated convolutions and channel-adaptive fusion strategies to enable flexible, scale-aware feature interaction. Extensive experiments on the challenging AI-TOD dataset show that the RDF-YOLO outperforms state-of-the-art methods by substantial margins. In particular, the RDF-YOLO improves AP50:95 by 6.9% and AP50by 8.6% over the YOLOv11 baseline and on small-object metrics, including APvt, APt, and APs. These results verify the effectiveness of the RDF-YOLO architecture for robust and efficient detection of small objects in remote sensing imagery. The source code is available at https://github.com/AssiiKk/RDF-YOLO.

Nowadays, application providers often prefer to execute their workflows on heterogeneous distributed computing resources deployed on cloud infrastructure to achieve a high level of resilience and cost saving. Optimally scheduling workflow on computing resources is a well-known combinatorial optimization problem, where a trend of using evolutionary algorithm (EA) is emerging rapidly. However, conventional EA optimizes only one problem in a single run and suffers from a high computational burden. In practical scenario, cloud platform needs to handle massive amounts of scheduling requests from users, scheduling different workflows simultaneously is highly challenging. Bearing this in mind, we put forward a novel knowledge transfer enabled EA to schedule diverse workflows in tandem, where domain knowledge of scheduling one workflow is extracted to enhance the scheduling efficiency of other related workflows. In our design, the knowledge source selection and the intensity of performing knowledge transfer are adapted in a synergistic way. Furthermore, search operator is enhanced by exploiting both historical experience and heuristic information. Experimental results on real-life workflows and extensive synthetic applications demonstrate the competitiveness of our approach, in comparison to state-of-the-art contenders. Note to Practitioners—Workflow scheduling is an important requirement for users in cloud computing, whose intractability increases exponentially when the size of problem grows, posing stiff challenges to heuristic methods. Using EAs to tackle workflow scheduling has received increasing attention recently. Suppose workflow scheduling is treated as a optimization task, cloud platform typically needs to handle versatile tasks from numerous users. However, traditional EA optimizes only one task in a single run and unable to handle multiple tasks at the same time. To address this issue, we introduce a novel multi-task solver to resolve different tasks jointly via online learning and exploitation of problem-solving experiences across tasks. The results demonstrate that our proposal significantly outperforms the state-of-the-art peers. It is expected to facilitate the practical efficacy of industrial cloud system which faces multiple workflow scheduling tasks submitted from enormous users.

Handling conflicting objectives and finding multiple Pareto optimal solutions are two challenging issues in solving multiobjective optimization problems (MOPs). Inspired by the efficiency of multitask optimization (MTO) in finding multiple optimal solutions of multitask optimization problem (MTOP), we propose to treat MOP as a MTOP and solve it by using MTO. By transforming the MOP into a MTOP, not only that the difficulty in handling conflicting objectives can be avoided, but also that MTO can help efficiently find well-distributed multiple optimal solutions for MOP. With the above idea, this paper proposes a new multiobjective optimization method via MTO, with the following three contributions. Firstly, a theorem is proposed to theoretically show the relationship between MOP and MTOP and how MOP can be transformed into a MTOP. Secondly, based on the theoretical analysis, a multiple tasks for multiple objectives (MTMO) framework is proposed for solving MOP efficiently. Thirdly, a MTMO-based evolutionary algorithm is developed to solve MOP, together with two novel strategies. One is a target point estimation strategy for transforming the MOP into a MTOP automatically and accurately. The other is an archive-based implicit knowledge transfer strategy for efficiently transferring knowledge across multiple tasks to enhance the optimization results of multiple tasks together. The superiority of the proposed algorithm is validated in extensive experiments on 15 MOPs with objective numbers varying from 3 to 20 and with six state-of-the-art algorithms as competitors. Therefore, solving MOP and even many-objective optimization problem via MTO is a new, promising, and efficient method.

With extensive pretrained knowledge and high-level general capabilities, large language models (LLMs) emerge as a promising avenue to augment reinforcement learning (RL) in aspects, such as multitask learning, sample efficiency, and high-level task planning. In this survey, we provide a comprehensive review of the existing literature in LLM-enhanced RL and summarize its characteristics compared with conventional RL methods, aiming to clarify the research scope and directions for future studies. Utilizing the classical agent-environment interaction paradigm, we propose a structured taxonomy to systematically categorize LLMs' functionalities in RL, including four roles: information processor, reward designer, decision-maker, and generator. For each role, we summarize the methodologies, analyze the specific RL challenges that are mitigated and provide insights into future directions. Finally, the comparative analysis of each role, potential applications, prospective opportunities, and challenges of the LLM-enhanced RL are discussed. By proposing this taxonomy, we aim to provide a framework for researchers to effectively leverage LLMs in the RL field, potentially accelerating RL applications in complex applications, such as robotics, autonomous driving, and energy systems.

Recent advances in machine learning (ML) for automating analog circuit synthesis have been significant, yet challenges remain. A critical gap is the lack of a standardized evaluation framework, compounded by various process design kits (PDKs), simulation tools, and a limited variety of circuit topologies. These factors hinder direct comparisons and the validation of algorithms. To address these shortcomings, we introduced AnalogGym, an open-source testing suite designed to provide fair and comprehensive evaluations. AnalogGym includes 30 circuit topologies in five categories: sensing front ends, voltage references, low dropout regulators, amplifiers, and phase-locked loops. It supports several technology nodes for academic and commercial applications and is compatible with commercial simulators such as Cadence Spectre, Synopsys HSPICE, and the open-source simulator Ngspice. AnalogGym standardizes the assessment of ML algorithms in analog circuit synthesis and promotes reproducibility with its open datasets and detailed benchmark specifications. AnalogGym's user-friendly design allows researchers to easily adapt it for robust, transparent comparisons of state-of-the-art methods, while also exposing them to real-world industrial design challenges, enhancing the practical relevance of their work. Additionally, we have conducted a comprehensive comparison study of various analog sizing methods on AnalogGym, highlighting the capabilities and advantages of different approaches. AnalogGym is available in the GitHub repository1. The documentations are also available at2.

Recent advances in convolutional neural networks (CNNs) have positioned them as leading models for linearizing power amplifiers (PAs) in digital predistortion (DPD) applications. However, current approaches often rely on a manually crafted architecture, leading to issues such as limited generalization ability and increased computational complexity. To address these issues, this study develops a method of growing neural architecture search (GNAS) for complex-valued CNN, or GNAS-CVCNN for short. A complex-valued convolution layer alongside a respective activation layer deals with the memory effects and nonlinear distortion of the PA and leverages on the properties of complex-valued signals prevalent in intermediate frequency (IF) domains. The GNAS-CVCNN optimally evolves the network structure and its associated parameters in tandem. This ensures the delivery of high performance with computational efficiency. We apply the GNAS-CVCNN to the linearization of two distinct real-world PAs whose operating bandwidth are 100MHz and 180MHz, respectively. Results comparison with the latest neural networks available reveals the superior linearization ability of the GNAS-CVCNN. Additionally, it achieves this higher performance with a reduced network size and enhanced processing speed, underscoring its practical efficacy in DPD applications.

With the proliferation of artificial intelligence, Computer-Aided Design (CAD) is being transformed into Computer-Automated Design (CAutoD). In this paper, the advent of Large Language Models (LLMs) introduces new opportunities for CAutoD. This study develops AutoForma, an LLM-based multi-agent system, for automatic conversion from natural language descriptions to 3D models. By harnessing the comprehension capabilities of LLMs, AutoForma streamlines the CAutoD workflow by efficiently translating design intents into precise models in CAD. Through a comprehensive set of evaluations, AutoForma is seen to offer automation performance across various design tasks, particularly in generating non-standard parts that meet specific requirements, with higher efficiency and accuracy than using just an LLM like GPT-4.

Data-efficient image classification poses a challenge in achieving effectiveness with limited data, as evidenced by the current methods based on convolutional neural networks (CNNs) and genetic programming (GP). Existing works employing these two methods encounter limitations, such as a lack of flexibility and an inability to effectively explore the latent features of the data. To tackle these challenges, this paper introduces a genetic programming method for data-efficient image recognition, leveraging novel function sets, terminal sets, and program structures. This method transforms tree-based data structures in GP into trainable CNN architectures. Further, by employing block structures instead of single operations in the search space, the search space is reduced and the stability of the search structures enhanced. Comparative experiments with state-of-the-art neural network methods and GP-based methods on data-efficient classification datasets validate the GP-CNN method offering higher performance.

Process, voltage, and temperature (PVT) variations in chip fabrication or operation pose a significant challenge to the robustness of analog integrated circuits. Existing design techniques for mitigating PVT variations involve analyzing offsets of DC operating points, but this approach often leads to compromises in circuit performance. To address this challenge, we developed a ‘PVT-Transfer’ framework to facilitate knowledge transfer with evolutionary design. Specifically, by cross-operating the circuit parameters under variations, design knowledge is transferred through parameter migration, thus enhancing the robustness of the resultant circuit. In addition, we leverage data-driven learning to discover potential similarities among PVT variations, thereby mitigating negative knowledge transfer. The PVT-Transfer Framework is evaluated on three integrated voltage references and compared with four state-of-the-art circuit sizing methods. Based on post-layout Monte-Carlo simulations, this framework is verified to offer superior performance to existing methods, yielding a 60% reduction in power consumption, an 80% increase in temperature resilience, and up to 70$\times$enhancement in the figure of merit. Further, it leads to a 60% reduction in the number of required circuit simulations and is suitable for parallel computation.

Semisupervised node classification is a prevalent task on graphs, which involves predicting the labels of unlabeled nodes based on limited labeled data available. At present, centralized approaches to training models for this task are unsustainable due to the increasing demand for computational power, storage capacity, and privacy. An approach of potential is federated graph learning (FGL), which allows multiple clients to collaborate on learning a model while maintaining data privacy. However, current methods suffer from the inability to consider the topology of the graph data and inadequate use of unlabeled data. To address these issues, we propose federated graph augmentation (FedGA) by combining graph neural network (GNN) models to utilize similar topologies existing in different client graphs and augment the client data. Furthermore, we develop FedGA-L based on FedGA, which integrates pseudolabeling and label-injection to improve the utilization of unlabeled data. FedGA-L allows pseudolabels to be used as additional information to enhance data augmentation and further improve the accuracy of node classification. We evaluate the effectiveness of FedGA and FedGA-L through experiments on multiple datasets. The results demonstrate improved accuracy in solving typical classification tasks and their compatibility with a variety of federated learning (FL) frameworks. On widely recognized datasets for graph learning, we achieve an accuracy improvement of 5%–7% compared to vanilla federated learning algorithms.

Cloud workflow scheduling (CWS) is critical for meeting user's high performance expectations in large-scale data processing and computing applications. CWS is known to be NP-hard and needs advanced scheduling techniques. Evolutionary algorithm and heuristic-based search techniques have gained massive popularity in addressing CWS, yet they either suffer from expensive computational cost or heavily rely on domain-specific experiences, which limit their practical applications. Bearing this in mind, we develop a novel evolutionary multi-task optimization framework to tackle a group of constrained CWS tasks simultaneously with the aid of adaptive cross-task problem-solving knowledge transfer. In particular, two collaborative knowledge exchange strategies, namely, constraint-free archive strategy and cross-task evolution strategy, are devised to extract useful building blocks from foreign tasks to boost the search efficiency. Further, to leverage the cooperative effects of both strategies, we develop an adaptive switching mechanism such that appropriate knowledge transfer strategies are learned automatically according to the population evolution status. Extensive experiments are conducted on real-world applications under various conditions, the comparison results show that our proposal delivers higher quality schedules than the state-of-the-art competitors in most cases.

Traffic signal timing optimization (TSTO) is a significant topic in the smart city. However, there are two challenges when solving TSTO. Firstly, it often uses time-consuming simulation software to evaluate candidate solutions, therefore it is an expensive optimization problem. Secondly, as the traffic flow changes rapidly in TSTO, providing a timing scheme to respond to the change immediately is difficult. To address the above challenges, we propose a region-based evaluation particle swarm optimization algorithm (REPSO) with dual solution libraries, which has three novel designs. First, two solution libraries are built for undersaturated and oversaturated traffic flow states, respectively, which can be used to fast provide a signal timing scheme for a traffic flow in real-time. Second, a knowledge-assisted initialization strategy is proposed and adopted to assist the initialization of new solutions based on the knowledge in the two solution libraries. Third, a region-based evaluation strategy is proposed to reduce the number of fitness evaluations, which can also greatly reduce the construction time of the solution libraries. The performance of REPSO is validated by comparing with six signal timing methods in both undersaturated and oversaturated traffic flow states, showing the better general performance of REPSO.

Designing analog integrated circuits (ICs), particularly sensors and reference circuits, requires a significant amount of human expertise and time, largely due to the requirement of maintaining process, voltage, and temperature (PVT) consistency. So far, there has been plenty of work on tuning the circuit to meet the PVT consistency requirements by comparing the offset of the DC operating point, but this inevitably leads to circuit performance degradation. To improve, we propose a ‘PVT-Transfer’ framework that utilizes knowledge transfer among PVT corners through evolutionary multitasking. Specifically, via cross-operating the circuit parameters under different PVT corners, knowledge is transferred through parameter migration to improve the robustness of the circuit. Further, PVT-Transfer employs data-driven learning to identify potential similarities among PVT variations, thereby leading to more cost-effective optimization. This framework is evaluated on two voltage references and compared with four state-of-the-art circuit sizing methods. The post-layout Monte-Carlo simulation results verify that PVT-Transfer outperforms the existing methods. It reduces the number of simulations required by 60% compared to the GCN-RL method. Besides, PVT-transfer achieves up to 10× improvement in the figure of merit over the human design.

Large-scale optimization problems (LSOPs) are challenging because the algorithm is difficult in balancing too many dimensions and in escaping from trapped bottleneck dimensions. To improve solutions, this paper introduces targeted modification to the certain values in the bottleneck dimensions. Analogous to gene targeting (GT) in biotechnology, we experiment on targeting the specific genes in candidate solution to improve its trait in differential evolution (DE). We propose a simple and efficient method, called GT-based DE (GTDE), to solve LSOPs. In the algorithm design, a simple GT-based modification is developed to perform on the best individual, comprising probabilistically targeting the location of bottleneck dimensions, constructing a homologous targeting vector, and inserting the targeting vector into the best individual. In this way, all the bottleneck dimensions of the best individual can be probabilistically targeted and modified to break the bottleneck and to provide global guidance for more optimal evolution. Note that the GT is only performed on the globally best individual and is just carried out as a simple operator that is added to the standard DE. Experimental studies compare the GTDE with some other state-of-the-art large-scale optimization algorithms, including the winners of CEC2010, CEC2012, CEC2013, and CEC2018 competitions on large-scale optimization. The results show that the GTDE is efficient and performs better than or at least comparable to the others in solving LSOPs.

Numerical comparison serves as a major tool in evaluating the performance of optimization algorithms, especially nondeterministic algorithms, but existing methods may suffer from a ‘cycle ranking’ paradox and/or a ‘survival of the non-fittest’ paradox. This paper searches for paradox-free data analysis methods for numerical comparison. It is discovered that a class of sufficient conditions exist for designing paradox-free analysis. Rigorous modeling and deduction are applied to a class of profile methods employing a filter. It is thus further discovered and proven that algorithm-independent filter conditions can prevent cycle ranking and survival of non-fittest paradoxes from occurring. By adopting an algorithm-independent filter, popular profile methods such as the ‘modified data profile method’‘, the accuracy profile method’, and ‘the operational characteristics zones method’ can be paradox free in comparing or benchmarking the performance of optimization algorithms.

Disassembly is the first phase to demanufacture end-of-life (EOL) products that are separated into parts/components for recovery. The quality conditions of EOL products are highly uncertain, which would result in some uncertain information during the disassembly process, e.g., the disassembly time and recovering revenue of each subassembly. It is quite challenging to determine the optimal/near-optimal disassembly solutions under uncertain information. This article studies uncertain disassembly planning (UDP) and proposes a self-adaptive learning approach to quickly identify the near-optimal disassembly solutions. First, we model the UDP by extending Petri nets, where not only disassembly operations but also EOL options of each subassembly are represented in the extended Petri Net. Based on the UDP model, we develop the self-adaptive learning approach, which integrates an approximation procedure for estimating uncertain disassembly information, aQ-learning algorithm for training disassembly samples, and a heuristic method for selecting the best disassembly solution. Finally, a hybrid Li-ion battery pack of Audi A3 Sportback e-tron is selected as the case study and applied to test the proposed self-adaptive learning approach. The experimental results demonstrate that our proposed method can efficiently find a better disassembly solution than the existing disassembly solution within 200 trainings in the case study.

In recent years, themulti-task learning forknowledge graph-basedrecommender system, termed MKR, has shown its promising performance and has attracted increasing interest, because a recommendation task and a knowledge graph embedding (KGE) task can help each other to improve the recommendation. However, MKR still has two difficult issues. The first is how fully to capture users’ historical behavior pattern in the recommendation task and how fully to utilize deep multi-relation semantic information in the KGE task. The second is how to deal with datasets with different sparsity. Tackling these challenging issues, this paper proposes an enhanced MKR (EMKR) approach with two novelties. First, we propose to utilize the attention mechanism to aggregate users’ historical behavior for more accurately mining preferences in the recommendation task, and utilize the relation-aware graph convolutional neural network to fully capture the deep multi-relation neighborhood features in the KGE task, so as to address the first issue. Second, a two-part modeling strategy is proposed for a better representation of users in the recommendation task to expand the expressive ability of the model for adapting to datasets with different sparsity, so as to address the second issue. Extensive experiments are conducted on widely-used datasets and 11 approaches are used for comparison. The results show that the proposed EMKR can achieve substantial gains over the compared state-of-the-art approaches, especially in the situation where user-item interactions are sparse.

Multimodal optimization problems (MMOPs) are sorts of optimization problems that have many global optima. To discover as many peaks as possible and increase the accuracy of the solutions, MMOP requires algorithms with great exploration and exploitation abilities. However, exploration and exploitation are in an adversarial relationship, since exploration aims to locate more optima via searching the global space rather than small regions, whereas exploitation targets to enhance the accuracy of solutions via searching in small areas. The key to efficiently solving MMOPs lies in striking a balance between exploration and exploitation. To achieve the goal, this paper proposes an adversarial differential evolution (ADE), containing an adversarial reproduction strategy and an adversarial selection strategy. Firstly, adversarial reproduction strategy generates offspring for exploration and offspring for exploitation and lets these two types of offspring compete for survival. Secondly, adversarial selection strategy employs a diversity-optimization-based selection and a crowding-based selection to select the offspring with both good diversity and good fitness. Diversity-optimization-based selection transforms the problem of selecting diverse individuals into an optimizing problem and solves it via an extra genetic algorithm to get the offspring with optimal diversity. Extensive experiments are conducted on CEC2013 MMOP benchmark to verify the effectiveness and efficiency of the proposed ADE. Experimental results show that ADE has advantages over the state-of-the-art MMOP algorithms.

Surrogate-assisted evolutionary algorithms (SAEAs) have performed well on low- and medium-scale expensive optimization problems (EOPs). However, with the dimensionality increasing, existing SAEAs have trouble getting reliable surrogates for solving large-scale EOPs. In this paper, we propose a progressive sampling surrogate-assisted particle swarm optimization (PS-SAPSO) to efficiently solve the large-scale EOPs from the perspective of data collection and model training. For the data collection, a progressive sampling strategy with restart operation is proposed to collect the new sample solutions during the evolution process for training the radial basis function network (RBFN) surrogate. Specifically, a social learning particle swarm optimization is employed to generate the new sample solutions under the control of a progressively varying stop criterion. For the model training, a dynamic tunning strategy is proposed to obtain a reliable RBFN by adaptively adjusting the hyperparameter setting during the evolution process. The experimental result shows that PS-SAPSO can achieve competitive or better performance compared with four state-of-the-art SAEAs on widely used benchmark functions. Moreover, ablation experiments are conducted to show the effectiveness of the components of the PS-SAPSO algorithm.

Data-driven evolutionary algorithms (DDEAs) have shown strong capacity in solving optimization problems with the help of surrogate models. However, current DDEAs often fall into the trap of inaccurate surrogates if no additional data can be obtained for the surrogates update during the optimization process. In addition, when dealing with multiobjective optimization problems, the surrogates in DDEAs will further suffer the difficulty of aggravating cumulative predicted fitness error. To overcome these difficulties, we propose a two-surrogate collaboration (TSC) management method, in which a Kriging surrogate and a radial basis function network (RBFN) surrogate are adopted to approximate the real fitness evaluation for the parent solutions and the newly generated offspring solutions, respectively. This TSC management method can effectively reduce the prediction uncertainty and improve the prediction performance of each surrogate without any other surrogate update process. During the optimization process, we use a modified social learning particle swarm optimization (SLPSO) as the basic search method and propose an offline DDEA. With the help of TSC, our resulting SLPSO-TSC algorithm can search for potential optimal solutions quickly and effectively without the help of any real fitness evaluation during the optimization process. The performance of the proposed SLPSO-TSC algorithm is verified on eleven widely used benchmark problems and compared with five different offline DDEAs. The experimental results show the high competitiveness of our proposed SLPSO-TSC for offline data-driven multiobjective optimization.

Cold chain logistics (CCL) scheduling is important for smart cities as it directly affects the service quality and operating profits of logistics companies. However, traditional CCL models seldom reflect the real transportation environment, making the solutions hardly applicable to the real CCL scenes. Hence, this paper attempts to establish a multisource data-associated CCL model oriented to the real transportation environment. This environment is considered by employing the real-captured driving duration and distance between any two places. Three scheduling objectives (namely, quality losses, personnel and vehicle costs, and transportation costs) are taken into account. To efficiently solve the proposed multisource data-associated multiobjective CCL model, a multiple population-based multiobjective ant colony system (MPMOACS) approach is proposed. Based on the multiple populations for multiple objectives framework, the MPMOACS approach can optimize multiple objectives sufficiently, and thus obtain promising solutions distributed along the entire Pareto front. To further enhance the performance of the MPMOACS, a ranking-based local search strategy is also designed. Experiments are conducted on not only the existing benchmark instances but also a real environment-aware multisource dataset that is built based on real-captured transportation data of Guangzhou and Shenzhen, China. Compared with six state-of-the-art and very recent well-performing multiobjective optimization approaches, the proposed MPMOACS approach exhibits the overall best performance.

Advancements in digitalisation and cyber-physical automation are opening up new frontiers in the marine industry such as autonomous shipping and smart manufacturing of ships. However, the progress in automating the ship design process which is still very human dependent, is slow. This paper introduces an automated ship design and optimisation concept by Hybridising Evolutionary Algorithm and Morphing (HEAM) to enable a more intelligent and efficient ship design process. It maps the entire ship hull form into a genotype using Non-dominated Sorting Genetic Algorithm II (NSGA-II) and flexible modification through morphing to create efficient designs for the hull form. This new method achieves predictive designs with a performance improvement of 11.67%, while achieving increased efficiency and minimum user dependency.

Numerical comparison is often key to verifying the performance of optimization algorithms, especially, global optimization algorithms. However, studies have so far neglected issues concerning comparison strategies necessary to rank optimization algorithms properly. To fill this gap for the first time, we combine voting theory and numerical comparison research areas, which have been disjoint so far, and thus extend the results of the former to the latter for optimization algorithms. In particular, we investigate compatibility issues arising from comparing two and more than two algorithms, termed “C2” and “C2+” in this article, respectively. Through defining and modeling “C2” and “C2+” mathematically, it is uncovered and illustrated that numerical comparison can be incompatible. Further, two possible paradoxes, namely, “cycle ranking” and “survival of the nonfittest,” are discovered and analyzed rigorously. The occurrence probabilities of these two paradoxes are also calculated under the no-free-lunch assumption, which shows the first justifiable use of the impartial culture assumption from voting theory, providing a point of reference to the frequency of the paradoxes occurring. It is also shown that significant influence on these probabilities comes from the number of algorithms and the number of optimization problems studied in the comparison. Further, various limiting probabilities when the number of optimization problems goes to infinity are also derived and characterized. The results would help guide benchmarking and developing optimization and machine learning algorithms.

As the marine industry moves towards the industry 4.0 era, the role of automated smart design is becoming increasingly significant. This offers an ability to produce highly customisable design and to integrate with the product-lifecycle process such as digitalised ship production and ship operations to in an efficient process. Currently, the hull form optimisation process is performed manually using ‘trial-and-error’ approach, which is not efficient. Focusing on automated smart design, this paper introduces a hybrid evolutionary algorithm and morphing (HEAM). It works by mapping the entire hull form (phenotype) into a chromosome (genotype), which allows global shape modification using a novel 2D morphing method. By combining this 2D morphing and Genetic Algorithm (GA), it enables optimal hull designs to be produced more rapidly with no user intervention.

Virtual machine placement (VMP) and energy efficiency are significant topics in cloud computing research. In this paper, evolutionary computing is applied to VMP to minimize the number of active physical servers, so as to schedule underutilized servers to save energy. Inspired by the promising performance of the ant colony system (ACS) algorithm for combinatorial problems, an ACS-based approach is developed to achieve the VMP goal. Coupled with order exchange and migration (OEM) local search techniques, the resultant algorithm is termed an OEMACS. It effectively minimizes the number of active servers used for the assignment of virtual machines (VMs) from a global optimization perspective through a novel strategy for pheromone deposition which guides the artificial ants toward promising solutions that group candidate VMs together. The OEMACS is applied to a variety of VMP problems with differing VM sizes in cloud environments of homogenous and heterogeneous servers. The results show that the OEMACS generally outperforms conventional heuristic and other evolutionary-based approaches, especially on VMP with bottleneck resource characteristics, and offers significant savings of energy and more efficient use of different resources.

In pedagogy, teachers usually separate mixed-level students into different levels, treat them differently and teach them in accordance with their cognitive and learning abilities. Inspired from this idea, we consider particles in the swarm as mixed-level students and propose a level-based learning swarm optimizer (LLSO) to settle large-scale optimization, which is still considerably challenging in evolutionary computation. At first, a level-based learning strategy is introduced, which separates particles into a number of levels according to their fitness values and treats particles in different levels differently. Then, a new exemplar selection strategy is designed to randomly select two predominant particles from two different higher levels in the current swarm to guide the learning of particles. The cooperation between these two strategies could afford great diversity enhancement for the optimizer. Further, the exploration and exploitation abilities of the optimizer are analyzed both theoretically and empirically in comparison with two popular particle swarm optimizers. Extensive comparisons with several state-of-the-art algorithms on two widely used sets of large-scale benchmark functions confirm the competitive performance of the proposed optimizer in both solution quality and computational efficiency. Finally, comparison experiments on problems with dimensionality increasing from 200 to 2000 further substantiate the good scalability of the developed optimizer.

Particle filter track-before-detect (PF-TBD) algorithms offer improvements over track-after-detect algorithms in detecting and tracking dim targets. However, it suffers from the particle collapsing problem, which can lead to deteriorated detection and tracking performance. To address this issue, a Lamarckian particle filter track-before-detect (LPF-TBD) algorithm is developed in this paper. In the LPF-TBD, before a TBD resampling process, a particle update strategy is applied, which is based on Lamarckian overriding and elitist operators designed to improve the particle diversity and efficiency. The effectiveness of the LPF-TBD algorithm is demonstrated using a widely adopted experiment on a target with a low signal-to-noise ratio in an image sequence. Compared with the currently-popular multinomial resampling PF-TBD method, the posterior distribution in the LPF-TBD can be more sufficiently approximated by the particles. Test results show that the LPF-TBD offers higher detection and tracking performance, while strengthening the algorithmic efficiency of particle filtering and evolutionary algorithms.

Large-scale optimization has become a significant yet challenging area in evolutionary computation. To solve this problem, this paper proposes a novel segment-based predominant learning swarm optimizer (SPLSO) swarm optimizer through letting several predominant particles guide the learning of a particle. First, a segment-based learning strategy is proposed to randomly divide the whole dimensions into segments. During update, variables in different segments are evolved by learning from different exemplars while the ones in the same segment are evolved by the same exemplar. Second, to accelerate search speed and enhance search diversity, a predominant learning strategy is also proposed, which lets several predominant particles guide the update of a particle with each predominant particle responsible for one segment of dimensions. By combining these two learning strategies together, SPLSO evolves all dimensions simultaneously and possesses competitive exploration and exploitation abilities. Extensive experiments are conducted on two large-scale benchmark function sets to investigate the influence of each algorithmic component and comparisons with several state-of-the-art meta-heuristic algorithms dealing with large-scale problems demonstrate the competitive efficiency and effectiveness of the proposed optimizer. Further the scalability of the optimizer to solve problems with dimensionality up to 2000 is also verified.

Taking the advantage of estimation of distribution algorithms (EDAs) in preserving high diversity, this paper proposes a multimodal EDA. Integrated with clustering strategies for crowding and speciation, two versions of this algorithm are developed, which operate at the niche level. Then these two algorithms are equipped with three distinctive techniques: 1) a dynamic cluster sizing strategy; 2) an alternative utilization of Gaussian and Cauchy distributions to generate offspring; and 3) an adaptive local search. The dynamic cluster sizing affords a potential balance between exploration and exploitation and reduces the sensitivity to the cluster size in the niching methods. Taking advantages of Gaussian and Cauchy distributions, we generate the offspring at the niche level through alternatively using these two distributions. Such utilization can also potentially offer a balance between exploration and exploitation. Further, solution accuracy is enhanced through a new local search scheme probabilistically conducted around seeds of niches with probabilities determined self-adaptively according to fitness values of these seeds. Extensive experiments conducted on 20 benchmark multimodal problems confirm that both algorithms can achieve competitive performance compared with several state-of-the-art multimodal algorithms, which is supported by nonparametric tests. Especially, the proposed algorithms are very promising for complex problems with many local optima.

Detecting community structure has become one important technique for studying complex networks. Although many community detection algorithms have been proposed, most of them focus on separated communities, where each node can belong to only one community. However, in many real-world networks, communities are often overlapped with each other. Developing overlapping community detection algorithms thus becomes necessary. Along this avenue, this paper proposes a maximal clique based multiobjective evolutionary algorithm (MOEA) for overlapping community detection. In this algorithm, a new representation scheme based on the introduced maximal-clique graph is presented. Since the maximal-clique graph is defined by using a set of maximal cliques of original graph as nodes and two maximal cliques are allowed to share the same nodes of the original graph, overlap is an intrinsic property of the maximal-clique graph. Attributing to this property, the new representation scheme allows MOEAs to handle the overlapping community detection problem in a way similar to that of the separated community detection, such that the optimization problems are simplified. As a result, the proposed algorithm could detect overlapping community structure with higher partition accuracy and lower computational cost when compared with the existing ones. The experiments on both synthetic and real-world networks validate the effectiveness and efficiency of the proposed algorithm.

Seeking multiple optima simultaneously, which multimodal optimization aims at, has attracted increasing attention but remains challenging. Taking advantage of ant colony optimization (ACO) algorithms in preserving high diversity, this paper intends to extend ACO algorithms to deal with multimodal optimization. First, combined with current niching methods, an adaptive multimodal continuous ACO algorithm is introduced. In this algorithm, an adaptive parameter adjustment is developed, which takes the difference among niches into consideration. Second, to accelerate convergence, a differential evolution mutation operator is alternatively utilized to build base vectors for ants to construct new solutions. Then, to enhance the exploitation, a local search scheme based on Gaussian distribution is self-adaptively performed around the seeds of niches. Together, the proposed algorithm affords a good balance between exploration and exploitation. Extensive experiments on 20 widely used benchmark multimodal functions are conducted to investigate the influence of each algorithmic component and results are compared with several state-of-the-art multimodal algorithms and winners of competitions on multimodal optimization. These comparisons demonstrate the competitive efficiency and effectiveness of the proposed algorithm, especially in dealing with complex problems with high numbers of local optima.

Gesture recognition has emerged recently as a promising application in our daily lives. Owing to low cost, prevalent availability, and structural simplicity, RFID shall become a popular technology for gesture recognition. However, the performance of existing RFID-based gesture recognition systems is constrained by unfavorable intrusiveness to users, requiring users to attach tags on their bodies. To overcome this, we propose GRfid, a novel device-free gesture recognition system based on phase information output by COTS RFID devices. Our work stems from the key insight that the RFID phase information is capable of capturing the spatial features of various gestures with low-cost commodity hardware. In GRfid, after data are collected by hardware, we process the data by a sequence of functional blocks, namely data preprocessing, gesture detection, profiles training, and gesture recognition, all of which are well-designed to achieve high performance in gesture recognition. We have implemented GRfid with a commercial RFID reader and multiple tags, and conducted extensive experiments in different scenarios to evaluate its performance. The results demonstrate that GRfid can achieve an average recognition accuracy of 96.5 and 92.8 percent in the identical-position and diverse-positions scenario, respectively. Moreover, experiment results show that GRfid is robust against environmental interference and tag orientations.

Existing differential evolution (DE) algorithms often face two challenges. The first is that the optimization performance is significantly affected by the ad hoc configurations of operators and parameters for different problems. The second is the long runtime for real-world problems whose fitness evaluations are often expensive. Aiming at solving these two problems, this paper develops a novel double-layered heterogeneous DE algorithm and realizes it in cloud computing distributed environment. In the first layer, different populations with various parameters and/or operators run concurrently and adaptively migrate to deliver robust solutions by making the best use of performance differences among multiple populations. In the second layer, a set of cloud virtual machines run in parallel to evaluate fitness of corresponding populations, reducing computational costs as offered by cloud. Experimental results on a set of benchmark problems with different search requirements and a case study with expensive design evaluations have shown that the proposed algorithm offers generally improved performance and reduced computational time, compared with not only conventional and a number of state-of-the-art DE variants, but also a number of other distributed DE and high-performing evolutionary algorithms. The speedup is significant especially on expensive problems, offering high potential in a broad range of real-world applications.

As the world becomes more connected and customer's `needs and wants' lean more towards bespoke products, the way they are designed and manufactured must adapt. Recognised as the main driver for the future of the manufacturing value chain, Industry 4.0 (i4) is rapidly gaining momentum worldwide. One of the key enablers of the i4 design model to achieve mass customisation manufactured at a mass production cost is the computational intelligence based Computer-Automated Design (CAutoD). This paper demonstrates how CAutoD realises the i4 concept for smart design of future ships and smart ships through-life. Following an overview of the i4 and CAutoD interface, a smart ship design technique is introduced to form an automated closed-loop approach to the entire ship design process. Then, key challenges and future directions on this roadmap are discussed. Lastly, a framework in which the concepts of morphing and free-form deformation are embedded into an evolutionary algorithm is developed to automate the design and optimisation process of the hull form.

The particle filter (PF) offers significant advantages over other nonlinear filters for non-Gaussian systems. However, it suffers from particle degeneracy and impoverishment, which can lead to deteriorated estimation performance. Through analyzing the filtering and Lamarckian evolution processes in this paper, we develop a Lamarckian PF (LPF), based on the assertion that specific characteristics of an organism are inheritable directly by its offspring. The LPF thus uses an overriding operator for the offspring to inherit traits such that the particle impoverishment problem is mitigated and filtering performance improved. Compared with the generic PF and the latest PF improved by differential evolution (DEPF), the LPF approximates posterior distribution more sufficiently. Meanwhile, the LPF weakens the impact of algorithmic parameters. Measured against the DEPF, the LPF reduces the complexity and improves the filtering speed and accuracy at the same time. Experimental results verify the validity and advantages of this new nonlinear filter.

Most internal combustion engines are built on compression or spark ignition, which is far from optimal and the problem of which is more than optimization. This paper first improves a genetic algorithm (GA) for such an application, aiming at the potential invention of a homogeneous charge microwave ignition (HCMI) engine. For an HCMI system, search for optimal emitters under the intrinsic constraints of resonant frequencies forms a coupled constraint optimization problem and poses an intractable challenge to the GA and virtual prototyping for the invention. A predefined GA (PGA) is then developed to handle appropriate frequency ranges for this problem so as to allow the parameters of the emitter, as well as its structure, to be optimized in an evolutionary process. The heuristic search is compared with the deterministic NM simplex and the nondeterministic conventional GA. Results show that while the NM and GA heuristics find an insufficient mode, the PGA often finds the global maximum, with a higher convergence rate and independent of the algorithm's initial settings. When the complexity of the problem increases with the number of variables, the PGA also delivers a robust performance while the NM and the GA yield divergent results. This application confirms the viability and power of evolutionary heuristics in inventing novel real-world solutions if properly adapted.

Following the first three industrial revolutions, Industry 4.0 (I4) aims at realizing mass customization at a mass production cost. Currently, however, there is a lack of smart analytics tools for achieving such a goal. This paper investigates this issues and then develops a predictive analytics framework integrating cloud computing, big data analysis, business informatics, communication technologies, and digital industrial production systems. Computational intelligence in the form of a self-organizing map (SOM) is used to manage relevant big data for feeding potential customer needs and wants to smart designs for targeted productivity and customized mass production. The selection of patterns from big data with SOM helps with clustering and with the selection of optimal attributes. A car customization case study shows that the SOM is able to assign new clusters when growing knowledge of customer needs and wants. The self-organizing tool offers a number of features suitable to smart design that is required in realizing Industry 4.0.

Industry 4.0 is characterized by smart manufacturing, implementation of Cyber Physical Systems (CPS) for production, i.e., embedded actuators and sensors, networks of microcomputers, and linking the machines to the value chain. It further considers the digital enhancement and reengineering of products. It is also characterized by highly differentiated customized products, and well-coordinated combination of products and services, and also the value added services with the actual product or service, and efficient supply chain. All these challenges require continuous innovation and learning, which is dependent on people and enterprise's capabilities. Appropriate management approaches can play a vital role in the development of dynamic capabilities, and effective learning and innovation climate. This paper aims at offering a viewpoint on best suitable management practices which can promote the climate of innovation and learning in the organization, and hence facilitate the business to match the pace of industry 4.0. This paper is one of the initial attempts to draw the attention towards the important role of management practices in industry 4.0, as most of the recent studies are discussing the technological aspect. This paper also suggests empirical and quantitative investigation on these management approaches in the context of industry 4.0.

Following analyzing existing challenges in addressing the balance between exploration and exploitation encountered by evolutionary algorithms, this paper develops a Genetic Algorithm with speciation (GASP). It first incorporates a novel encoding scheme and recombination method for a balanced genetic divergence when locating global optima in complex applications, such as structural and dynamic design of an artificial neural network (NN). GASP also addresses the problem of defining a measure and track population diversity whose NN structure is subjected to continual reorganization during the evolution process. Further, a novel approach to the neural network phenotype is developed, which maps it to a distinct genome with a variable length capable of fully representing the multilayer feed-forward NN structure. Using the concept generalized from linguistic complexity, the distance between strings can thus be derived from the single string and substring counts. The GASP is then applied to an NN design problem to forecast the energy consumption of a built environment. With the optimal NN structure, diversity is tracked and improved. The results show that the GASP succeeds in obtaining excellent accuracy and speed.

In this paper, Adaptive Neuro-Fuzzy Interference System (ANFIS) technique is used to develop models to predict two conditions commonly found in a Wireless Sensor Network's deployment; these conditions are failure due to (i) poorly deployed environment and (ii) human movements. ANFIS models are trained using parameters obtained from actual ZigBee PRO nodes' Neighbour Table experimented under the influence of associated network challenges. These parameters are Mean RSSI, Standard Deviation RSSI, Average Coefficient of Variation RSSI and Neighbour Table Connectivity. The individual and combined effects of parameters are investigated in-depth. Results showed the mean RSSI is a critical parameter and the combination of mean RSSI, ACV RSSI and NTC produced the best prediction results (~92%) for all ANFIS models.

Social learning in particle swarm optimization (PSO) helps collective efficiency, whereas individual reproduction in genetic algorithm (GA) facilitates global effectiveness. This observation recently leads to hybridizing PSO with GA for performance enhancement. However, existing work uses a mechanistic parallel superposition and research has shown that construction of superior exemplars in PSO is more effective. Hence, this paper first develops a new framework so as to organically hybridize PSO with another optimization technique for "learning." This leads to a generalized "learning PSO" paradigm, the *L-PSO. The paradigm is composed of two cascading layers, the first for exemplar generation and the second for particle updates as per a normal PSO algorithm. Using genetic evolution to breed promising exemplars for PSO, a specific novel *L-PSO algorithm is proposed in the paper, termed genetic learning PSO (GL-PSO). In particular, genetic operators are used to generate exemplars from which particles learn and, in turn, historical search information of particles provides guidance to the evolution of the exemplars. By performing crossover, mutation, and selection on the historical information of particles, the constructed exemplars are not only well diversified, but also high qualified. Under such guidance, the global search ability and search efficiency of PSO are both enhanced. The proposed GL-PSO is tested on 42 benchmark functions widely adopted in the literature. Experimental results verify the effectiveness, efficiency, robustness, and scalability of the GL-PSO.

This paper studies the optimization problem of topological active net (TAN), which is often seen in image segmentation and shape modeling. A TAN is a topological structure containing many nodes, whose positions must be optimized while a predefined topology needs to be maintained. TAN optimization is often time-consuming and even constructing a single solution is hard to do. Such a problem is usually approached by a "best improvement local search" (BILS) algorithm based on deterministic search (DS), which is inefficient because it spends too much efforts in nonpromising probing. In this paper, we propose the use of micro-differential evolution (DE) to replace DS in BILS for improved directional guidance. The resultant algorithm is termed deBILS. Its micro-population efficiently utilizes historical information for potentially promising search directions and hence improves efficiency in probing. Results show that deBILS can probe promising neighborhoods for each node of a TAN. Experimental tests verify that deBILS offers substantially higher search speed and solution quality not only than ordinary BILS, but also the genetic algorithm and scatter search algorithm.

Operating mode scheduling is crucial for the lifetime of wireless sensor networks (WSNs). However, the growing scale of networks has made such a scheduling problem more challenging, as existing set cover and evolutionary algorithms become unable to provide satisfactory efficiency due to the curse of dimensionality. In this paper, a Kuhn–Munkres (KM) parallel genetic algorithm is developed to solve the set cover problem and is applied to the lifetime maximization of large-scale WSNs. The proposed algorithm schedules the sensors into a number of disjoint complete cover sets and activates them in batch for energy conservation. It uses a divide-and-conquer strategy of dimensionality reduction, and the polynomial KM algorithm a are hence adopted to splice the feasible solutions obtained in each subarea to enhance the search efficiency substantially. To further improve global efficiency, a redundant-trend sensor schedule strategy was developed. Additionally, we meliorate the evaluation function through penalizing incomplete cover sets, which speeds up convergence. Eight types of experiments are conducted on a distributed platform to test and inform the effectiveness of the proposed algorithm. The results show that it offers promising performance in terms of the convergence rate, solution quality, and success rate.

Existing multiobjective evolutionary algorithms (MOEAs) tackle a multiobjective problem either as a whole or as several decomposed single-objective sub-problems. Though the problem decomposition approach generally converges faster through optimizing all the sub-problems simultaneously, there are two issues not fully addressed, i.e., distribution of solutions often depends on a priori problem decomposition, and the lack of population diversity among sub-problems. In this paper, a MOEA with double-level archives is developed. The algorithm takes advantages of both the multiobjective-problem-level and the sub-problem-level approaches by introducing two types of archives, i.e., the global archive and the sub-archive. In each generation, self-reproduction with the global archive and cross-reproduction between the global archive and sub-archives both breed new individuals. The global archive and sub-archives communicate through cross-reproduction, and are updated using the reproduced individuals. Such a framework thus retains fast convergence, and at the same time handles solution distribution along Pareto front (PF) with scalability. To test the performance of the proposed algorithm, experiments are conducted on both the widely used benchmarks and a set of truly disconnected problems. The results verify that, compared with state-of-the-art MOEAs, the proposed algorithm offers competitive advantages in distance to the PF, solution coverage, and search speed.

Utilizing cumulative correlation information already existing in an evolutionary process, this paper proposes a predictive approach to the reproduction mechanism of new individuals for differential evolution (DE) algorithms. DE uses a distributed model (DM) to generate new individuals, which is relatively explorative, whilst evolution strategy (ES) uses a centralized model (CM) to generate offspring, which through adaptation retains a convergence momentum. This paper adopts a key feature in the CM of a covariance matrix adaptation ES, the cumulatively learned evolution path (EP), to formulate a new evolutionary algorithm (EA) framework, termed DEEP, standing for DE with an EP. Without mechanistically combining two CM and DM based algorithms together, the DEEP framework offers advantages of both a DM and a CM and hence substantially enhances performance. Under this architecture, a self-adaptation mechanism can be built inherently in a DEEP algorithm, easing the task of predetermining algorithm control parameters. Two DEEP variants are developed and illustrated in the paper. Experiments on the CEC'13 test suites and two practical problems demonstrate that the DEEP algorithms offer promising results, compared with the original DEs and other relevant state-of-the-art EAs.

The University of Glasgow-University of Electronic Science and Technology of China joint education programme enrolled its first cohort of students in Fall 2013. Continual assessment of the programme was initiated shortly thereafter to monitor student achievement in the programme, to evaluate how well the aims of the programme are being met, and to determine areas where improvements can be made. Several issueswere identified during the review as well as in informal conversations with course instructors, administrative support staff, and students. To quantify the impact of these issues, a study to examine the cultural attitudes, English language communication aptitude, problem-solving skills, learning style, and teacher-student relationship of the first cohort during their first year in its undergraduate programme in electronics and electrical engineering has been proposed The study will adopt an interpretive approach, using a longitudinal multivariate study design, to generate quantified descriptions of how each issue influences student behavior and success in the programme. A follow-on study on the course instructors' cultural attitudes, pedagogical approaches, and willingness to adapt is planned. The initial findings from the programme assessment and the proposed studies will be presented at the 2014 Frontiers in Education Conference in Madrid, Spain in October 2014.

Among renewable generators, photovoltaics (PV) is showing an increasing suitability and a lowering cost. However, integration of renewable energy sources possesses many challenges, as the intermittency of these non-conventional sources often requires generation forecast, planning and optimal management. There exists scope to improve present PV yield forecasting models and methods. For example, the popular dynamic neural network modelling method suffers from the lack of a selection mechanism for an optimal network structure. This paper develops an enhanced network for short-term forecasting of PV power yield, termed a `focused time-delay neural network' (FTDNN). The problem of optimizing the FTDNN structure is reduced to optimizing the number of delay steps and the number of neurons in the hidden layer alone and this problem is conveniently solved through heuristics. Two such algorithms, a genetic algorithm and particle swarm optimization (PSO) have been tested and both prove efficient and can improve the forecasting accuracy of the dynamic network. Given the success of the PSO in solving this discontinuous structural optimization problem, it is expected that PSO offers potential in optimizing both the structure and parameters of a forecasting model.

Traditionally, the Evolutionary Computation (EC) paradigm is inspired by Darwinian evolution or the swarm intelligence of animals. Bandura's Social Learning Theory pointed out that the social learning behavior of humans indicates a high level of intelligence in nature. We found that such intelligence of human society can be implemented by numerical computing and be utilized in computational algorithms for solving optimization problems. In this paper, we design a novel and generic optimization approach that mimics the social learning process of humans. Emulating the observational learning and reinforcement behaviors, a virtual society deployed in the algorithm seeks the strongest behavioral patterns with the best outcome. This corresponds to searching for the best solution in solving optimization problems. Experimental studies in this paper showed the appealing search behavior of this human intelligence-inspired approach, which can reach the global optimum even in ill conditions. The effectiveness and high efficiency of the proposed algorithm has further been verified by comparing to some representative EC algorithms and variants on a set of benchmarks.

In nature, almost every organism ages and has a limited lifespan. Aging has been explored by biologists to be an important mechanism for maintaining diversity. In a social animal colony, aging makes the old leader of the colony become weak, providing opportunities for the other individuals to challenge the leadership position. Inspired by this natural phenomenon, this paper transplants the aging mechanism to particle swarm optimization (PSO) and proposes a PSO with an aging leader and challengers (ALC-PSO). ALC-PSO is designed to overcome the problem of premature convergence without significantly impairing the fast-converging feature of PSO. It is characterized by assigning the leader of the swarm with a growing age and a lifespan, and allowing the other individuals to challenge the leadership when the leader becomes aged. The lifespan of the leader is adaptively tuned according to the leader's leading power. If a leader shows strong leading power, it lives longer to attract the swarm toward better positions. Otherwise, if a leader fails to improve the swarm and gets old, new particles emerge to challenge and claim the leadership, which brings in diversity. In this way, the concept “aging” in ALC-PSO actually serves as a challenging mechanism for promoting a suitable leader to lead the swarm. The algorithm is experimentally validated on 17 benchmark functions. Its high performance is confirmed by comparing with eight popular PSO variants.

Railway timetable scheduling is a fundamental operational problem in the railway industry and has significant influence on the quality of service provided by the transport system. This paper explores the periodic railway timetable scheduling (PRTS) problem, with the objective to minimize the average waiting time of the transfer passengers. Unlike traditional PRTS models that only involve service lines with fixed cycles, this paper presents a more flexible model by allowing the cycle of service lines and the number of transfer passengers to vary with the time period. An enhanced differential evolution (DE) algorithm with dual populations, termed “dual-population DE” (DP-DE), was developed to solve the PRTS problem, yielding high-quality solutions. In the DP-DE, two populations cooperate during the evolution; the first focuses on global search by adopting parameter settings and operators that help maintain population diversity, while the second one focuses on speeding up convergence by adopting parameter settings and operators that are good for local fine tuning. A novel bidirectional migration operator is proposed to share the search experience between the two populations. The proposed DP-DE has been applied to optimize the timetable of the Guangzhou Metro system in Mainland China and six artificial periodic railway systems. Two conventional deterministic algorithms and seven highly regarded evolutionary algorithms are used for comparison. The comparison results reveal that the performance of DP-PE is very promising.

The importance of predicting the output power of Photovoltaic (PV) plants is crucial in modern power system applications. Predicting the power yield of a PV generation system helps the process of dispatching the power into a grid with improved efficiency in generation planning and operation. This work proposes the use of intelligent tools to forecast the real power output of PV units. These tools primarily comprise dynamic neural networks which are capable of time-series predictions with good reliability. This paper begins with a brief review of various methods of forecasting solar power reported in literature. Results of preliminary work on a 5kW PV panel at King Fahd University of Petroleum and Minerals, Dhahran, Saudi Arabia, is presented. Focused Time Delay and Distributed Time Delay Neural Networks were used as a forecasting tool for this study and their performance was compared with each other.

Developing techniques for optimal allocation of limited resources to a set of activities has received increasing attention in recent years. In this paper, an efficient resource allocation scheme based on particle swarm optimization (PSO) is developed. Different from many existing evolutionary algorithms for solving resource allocation problems (RAPs), this PSO algorithm incorporates a novel representation of each particle in the population and a comprehensive learning strategy for the PSO search process. The novelty of this representation lies in that the position of each particle is represented by a pair of points, one on each side of the constraint hyper-plane in the problem space. The line joining these two points intersects the constraint hyper-plane and their intersection point indicates a feasible solution. With the evaluation value of the feasible solution used as the fitness value of the particle, such a representation provides an effective way to ensure the equality resource constraints in RAPs are met. Without the distraction of infeasible solutions, the particle thus searches the space smoothly. In addition, particles search for optimal solutions by learning from themselves and their neighborhood using the comprehensive learning strategy, helping prevent premature convergence and improve the solution quality for multimodal problems. This new algorithm is shown to be applicable to both single-objective and multiobjective RAPs, with performance validated by a number of benchmarks and by a real-world bed capacity planning problem. Experimental results verify the effectiveness and efficiency of the proposed algorithm.

Maximizing the lifetime of wireless sensor networks (WSNs) is a challenging problem. Although some methods exist to address the problem in homogeneous WSNs, research on this problem in heterogeneous WSNs have progressed at a slow pace. Inspired by the promising performance of ant colony optimization (ACO) to solve combinatorial problems, this paper proposes an ACO-based approach that can maximize the lifetime of heterogeneous WSNs. The methodology is based on finding the maximum number of disjoint connected covers that satisfy both sensing coverage and network connectivity. A construction graph is designed with each vertex denoting the assignment of a device in a subset. Based on pheromone and heuristic information, the ants seek an optimal path on the construction graph to maximize the number of connected covers. The pheromone serves as a metaphor for the search experiences in building connected covers. The heuristic information is used to reflect the desirability of device assignments. A local search procedure is designed to further improve the search efficiency. The proposed approach has been applied to a variety of heterogeneous WSNs. The results show that the approach is effective and efficient in finding high-quality solutions for maximizing the lifetime of heterogeneous WSNs.

Particle swarm optimization (PSO) relies on its learning strategy to guide its search direction. Traditionally, each particle utilizes its historical best experience and its neighborhood's best experience through linear summation. Such a learning strategy is easy to use, but is inefficient when searching in complex problem spaces. Hence, designing learning strategies that can utilize previous search information (experience) more efficiently has become one of the most salient and active PSO research topics. In this paper, we proposes an orthogonal learning (OL) strategy for PSO to discover more useful information that lies in the above two experiences via orthogonal experimental design. We name this PSO as orthogonal learning particle swarm optimization (OLPSO). The OL strategy can guide particles to fly in better directions by constructing a much promising and efficient exemplar. The OL strategy can be applied to PSO with any topological structure. In this paper, it is applied to both global and local versions of PSO, yielding the OLPSO-G and OLPSO-L algorithms, respectively. This new learning strategy and the new algorithms are tested on a set of 16 benchmark functions, and are compared with other PSO algorithms and some state of the art evolutionary algorithms. The experimental results illustrate the effectiveness and efficiency of the proposed learning strategy and algorithms. The comparisons show that OLPSO significantly improves the performance of PSO, offering faster global convergence, higher solution quality, and stronger robustness.

The aircraft arrival sequencing and scheduling (ASS) problem is a salient problem in air traffic control (ATC), which proves to be nondeterministic polynomial (NP) hard. This paper formulates the ASS problem in the form of a permutation problem and proposes a new solution framework that makes the first attempt at using an ant colony system (ACS) algorithm based on the receding horizon control (RHC) to solve it. The resultant RHC-improved ACS algorithm for the ASS problem (termed the RHC-ACS-ASS algorithm) is robust, effective, and efficient, not only due to that the ACS algorithm has a strong global search ability and has been proven to be suitable for these kinds of NP-hard problems but also due to that the RHC technique can divide the problem with receding time windows to reduce the computational burden and enhance the solution's quality. The RHC-ACS-ASS algorithm is extensively tested on the cases from the literatures and the cases randomly generated. Comprehensive investigations are also made for the evaluation of the influences of ACS and RHC parameters on the performance of the algorithm. Moreover, the proposed algorithm is further enhanced by using a two-opt exchange heuristic local search. Experimental results verify that the proposed RHC-ACS-ASS algorithm generally outperforms ordinary ACS without using the RHC technique and genetic algorithms (GAs) in solving the ASS problems and offers high robustness, effectiveness, and efficiency.

An ant colony optimization (ACO) algorithm offers algorithmic techniques for optimization by simulating the foraging behavior of a group of ants to perform incremental solution constructions and to realize a pheromone laying-and-following mechanism. Although ACO is first designed for solving discrete (combinatorial) optimization problems, the ACO procedure is also applicable to continuous optimization. This paper presents a new way of extending ACO to solving continuous optimization problems by focusing on continuous variable sampling as a key to transforming ACO from discrete optimization to continuous optimization. The proposed SamACO algorithm consists of three major steps, i.e., the generation of candidate variable values for selection, the ants' solution construction, and the pheromone update process. The distinct characteristics of SamACO are the cooperation of a novel sampling method for discretizing the continuous search space and an efficient incremental solution construction method based on the sampled values. The performance of SamACO is tested using continuous numerical functions with unimodal and multimodal features. Compared with some state-of-the-art algorithms, including traditional ant-based algorithms and representative computational intelligence algorithms for continuous optimization, the performance of SamACO is seen competitive and promising.

An adaptive particle swarm optimization (APSO) that features better search efficiency than classical particle swarm optimization (PSO) is presented. More importantly, it can perform a global search over the entire search space with faster convergence speed. The APSO consists of two main steps. First, by evaluating the population distribution and particle fitness, a real-time evolutionary state estimation procedure is performed to identify one of the following four defined evolutionary states, including exploration, exploitation, convergence, and jumping out in each generation. It enables the automatic control of inertia weight, acceleration coefficients, and other algorithmic parameters at run time to improve the search efficiency and convergence speed. Then, an elitist learning strategy is performed when the evolutionary state is classified as convergence state. The strategy will act on the globally best particle to jump out of the likely local optima. The APSO has comprehensively been evaluated on 12 unimodal and multimodal benchmark functions. The effects of parameter adaptation and elitist learning will be studied. Results show that APSO substantially enhances the performance of the PSO paradigm in terms of convergence speed, global optimality, solution accuracy, and algorithm reliability. As APSO introduces two new parameters to the PSO paradigm only, it does not introduce an additional design or implementation complexity.

The collapse of the Californian electricity market system in 2001 has highlighted urgency in research in intelligent electricity trading systems and strategies involving both suppliers and customs. In their trading systems, power generation companies under the New Electricity Trading Arrangement (NETA) of the UK are now developing gaming strategies. However, modelling of such “intelligent” market behaviours is extremely challenging, because traditional mathematical and computer modelling techniques cannot cope with the involvement of game theory. In this paper, evolutionary computation enabled modelling of such system is presented. Both competitive and cooperative game theory strategies are taken into account in evolving the intelligent model. The model then leads to intelligent trading strategy development and decision support. Experimental tests, verification and validation are carried out with various strategies, using different model scales and data published by NETA. Results show that evolutionary computation enabled game theory involved modelling and decision making provides an effective tool for NETA trading analysis, prediction and support.

Although multi-objective evolutionary algorithm techniques are becoming mature, benchmark measures for evaluating the algorithms still require further research, as convergence theories can hardly be applied here and the only practical method for performance comparison is through benchmark tests. This paper investigates the current progress on multi-objective evolutionary algorithm performance measurement. The paper is focused on identifying deficiencies existing in the current performance measure techniques. It is shown that, whilst some performance indicators are conclusive and consistent, it is critical for some cases to include the 'diversity' indicator in a benchmark test. Possible ways forward are also identified.

The paper develops a genetic algorithm (GA) based "growing" technique to design and synthesise analogue circuits with practical constraints, such as the manufacturer's preferred component values. Most existing problems when evolutionary search techniques are applied to circuit design are addressed. The developed GA technique is then applied both to synthesise the topology of a network and perform value optimisation on the components based on a set of commonly used component values (E-12 series). Passive filter networks synthesised this way are realisable, effective and of novel topology. It is anticipated that this technique can be extended to active networks.

Simple, linear classical controllers are highly popular in industrial applications. However, most controllers have to be tuned and manually re-tuned on a trial and error basis at every operating level. This is particularly difficult when the plant to be controlled is significantly nonlinear. The deficiency in localised linearised models associated with 'local model networks' has been overcome by the introduction of 'linear approximation model (LAM) networks'. To address this problem and help in the design of industrial controllers for a wider range of operating trajectories,y, this paper develops a controller network design technique based upon a LAM network of a practical or nonlinear system to be controlled. This is called a 'Trajectory Controller Network (TCN)', which overcomes the deficiency associated with local controller networks. Each element of a TCN can be of a simple form, such as PID, and may be obtained directly from a set of step response data at several typical operating levels for fast prototyping. Since plant step response data are often readily available in control engineering practice, such TCNs can be automatically and optimally evolved from these data directly without the need for model identification. The overall controller is co-ordinated and evolved along the entire operating trajectory in the operating envelope, tackling the control problem of practical or nonlinear plants. Evolutionary computation provides global structural search for the network and multi-objective optimisation of the controllers. This novel technique is illustrated and validated through a nonlinear control example.